Search

Various morphological proxies have been used to infer habitat preferences among fossil turtles and their early ancestors, but most are tightly linked to phylogeny, thereby minimizing their predictive power. One particularly widely used model incorporates linear measurements of the forelimb (humerus + ulna + manus), but in addition to the issue of phylogenetic correlation, it does not estimate the likelihood of habitat assignment. Here, we introduce a new model that uses intramanual measurements (digit III metacarpal + non‐ungual phalanges + ungual) to statistically estimate habitat likelihood and that has greater predictive strength than prior estimators. Application of the model supports the hypothesis that stem‐turtles were primarily terrestrial in nature and recovers the nanhsiungchelyid Basilemys (a fossil crown‐group turtle) as having lived primarily on land, despite some prior claims to the contrary., Various morphological proxies have been used to infer habitat preferences among fossil turtles and their early ancestors, but most are tightly linked to phylogeny, thereby minimizing their predictive power. Here, we introduce a new model that uses intramanual measurements (digit III metacarpal + non‐ungual phalanges + ungual) to statistically estimate habitat likelihood and that has greater predictive strength than prior estimators. Application of the model supports the hypothesis that stem turtles were primarily terrestrial in nature and recovers the nanhsiungchelyid Basilemys (a fossil crown‐group turtle) as having lived primarily on land, despite some prior claims to the contrary.